Bayes.Script

Documentation

taskConcepts :: TaskID -> Evidence -> Evidence

Run the evidence on the Bayesian networks to add soft evidence.

Instead of running it on the combined model, we run it through the models seperately and only output the evidence as it should be in the student model nodes. This should work since there should be no feedback effects at that point. feed :: Evidence -> StudentModel feed = toEvidence StudentModel.network . feed'

Obtain task concept node probabilities for a particular task.

emptyConcepts :: [(TaskID, Evidence)]

Feeding empty evidence to the task concept evidence. This is used in case the task is not even mentioned in the precalculations (presumably because no request for the task was sent).

aggregateConcepts :: [(TaskID, Evidence)] -> Evidence

Aggregate task concept node probabilities into generic concept node probabilities.

aggregateConcepts' :: [(TaskID, Evidence)] -> Evidence

Same as aggregateConcepts, but do not include task concept node probabilities in final evidence.

findEvidence :: Monad m => XML -> m Evidence

addModelToLatestRequest :: Connection -> IO ()

Add information from the student modeller to the database, based on the raw input and output of the domain reasoner. This is unfortunately done in a roundabout way, due to the way the framework functions.

We assume that the latest row to the requests table is the one we're dealing with, since we must know the row ID to continue.